---
title: Get Started With Autotune
description: "Set up a Statsig Autotune experiment, including selecting a goal metric, defining variants, configuring traffic allocation, and launching to users."
product: general
lang: en
last_updated: 2025-09-18
token_estimate: 1040
---
# Get Started With Autotune

> For AI agents: a documentation index is available at [/llms.txt](/llms.txt). Append `.md` to any page URL for markdown, or send `Accept: text/markdown`.

Autotune adjusts traffic among variants to optimize a success event and select one winning variant for all users. Use it when you want one global winner. Use contextual Autotune to personalize variant selection based on each user's context features.

## How to set up Autotune

1. To create a new Autotune experiment, navigate to the [Autotune section on the Statsig console](https://console.statsig.com/autotune).
2. Click **Create**, enter a name and a goal for the Autotune experiment, and set its ID type. Optionally assign a team, target applications, and tags. Click **Create** again. Statsig opens the Autotune's _Setup_ tab, where you configure the rest.
3. In _Groups and Parameters_, provide the variants to test. Each variant needs a name, and a corresponding JSON value. Statsig returns the variant listed as Control/Default when the Autotune experiment isn't running.

![Autotune experiment variant configuration interface](https://docs.statsig.com/images/autotune/setup/131385189-5f0c1d93-ba87-4159-8995-3c30991587a0.png)

4. Select the success event to optimize for. You can further specify an optional [event value](https://docs.statsig.com/guides/logging-events).

![Autotune success event selection interface](https://docs.statsig.com/images/autotune/setup/131385239-5a76d253-022b-457e-a370-f9ee7ce566a1.png)

You can specify a few parameters:

| Parameter | Description |
| --- | --- |
| Measurement Experiment | An existing experiment whose results Statsig shows alongside this Autotune, so you can measure the Autotune experience against a fixed baseline. Selecting one doesn't split traffic or change evaluation: your code creates the comparison by checking that experiment and calling the Autotune in its treatment branch. Go to [Linked experiments](https://docs.statsig.com/autotune/monitoring#linked-experiments). |
| Exploration Window | The initial time period where Autotune equally splits traffic. This is useful for noisy or temporal metrics where hourly data swings can bias Autotune's initial measurements. |
| Attribution Window | The maximum duration between the exposure and success event that counts as a success. Statsig recommends 1 hour for most applications; adjust if you expect the success event to lag the exposure event by several hours. |
| Winner Threshold | The "probability of best" threshold a variant must reach for Autotune to declare it the winner, stop collecting data, and direct all traffic. A lower value results in faster decisions but increases the probability of selecting the wrong winner. |

If your attribution window is longer than your exploration window, Autotune continues to split traffic equally until the attribution window ends. Until then, Autotune doesn't have complete outcome data.

Click **Create** to finalize the setup.

6. As with Feature Gates and Experiments, Statsig provides a code snippet for the exposure check event to add to your code. Select **Start** when you're ready to launch your Autotune test.

![Autotune code snippet and launch interface](https://docs.statsig.com/images/autotune/setup/131384977-144dd868-787b-45ad-9ff1-fc9afbd4c769.png)

## Schedule the start time

Instead of launching an Autotune right away, you can schedule it to start at a specific time. This is useful when you want a test to begin during a particular window, such as after a holiday, without having to start it manually.

To schedule the start:

1. Select **Start** to open the start dialog.
2. Turn on **Schedule Autotune Start Time**.
3. Choose the date and time you want the Autotune to begin. You can schedule a start up to one year in advance, and times snap to the nearest 15-minute interval. The picker uses your local time zone.
4. Select **Schedule Start**.

The Autotune shows a Scheduled status until the start time, when it launches automatically. To change or remove a scheduled start before it begins, use the **Edit Scheduled Start** or **Cancel Scheduled Start** option on the Autotune.